This paper investigates the market microstructure of the Taiwan Stock Exchange Capitalization weighted Stock Index (TX) futures contracts traded on the Taiwan Futures Exchange which quite recently switched from an electronic periodic call auction market to an electronic continuous auction market. No doubt it is a rare opportunity for us to deeply look into market quality under different trading mechanisms. Using time-stamped transaction data of trades and quotes covering the period from January 2001 to September 2002, overall behavior for all TX Futures contracts are explored first—including intraday and daily patterns in the bid-ask spreads, volume, trade size, volatility, liquidity ratio and other characteristics. Next, in order to observe whether long-term contracts and short-term contracts have different patterns, the sample is divided into two groups—quarterly expiration contract months (March, June, September, and December) and non-quarterly expiration contract months, and the intraday/daily patterns are displayed. Moreover, since TAIFEX transferred trading mechanism on July 29th 2002 from an electronic periodic call auction market to an electronic continuous auction market, intraday/daily patterns are separately illustrated and compared before and after July 29th 2002, and ANOVA F-Statistic and Kruskal-Wallis tests are also taken to provide more insights into time-varying behavior under two different kinds of market trading mechanisms. The empirical results indicate that the most active periods correspond to the TAIFEX’Ss opening five-minute interval (8:45-8:50), TSEC’Ss opening five-minute interval (9:00-9:05), and TAIFEX’S closing five-minute interval (13:40-13:45) with wide spreads and large trade sizes. In 54 five-minute intervals for the regular trading session of both TAIFEX and TSEC from 9:00 a.m. to 1:30 p.m., the behavior of spreads, volume and trade sizes mainly reveal U-shaped patterns. The average trading volume within each time interval plunges, except within the final 5 minutes closing procedure interval, after TAIFEX transferred trading mechanism from an electronic periodic call auction market to an electronic continuous auction market with wider spreads and narrower volatility, in general. Moreover, intraday patterns of the average volume under new microstructure exhibit a right angular U-shape while intraday patterns of volume under old microstructure reveal a smooth U-shape. The evidence suggests a conjecture that the transfer of market trading mechanism might result in informed traders’ altering their intraday behavior and might lessening their trading desire. Further evidence in confirmation of this statement is left to future work. / This paper investigates the market microstructure of the Taiwan Stock Exchange Capitalization weighted Stock Index (TX) futures contracts traded on the Taiwan Futures Exchange which quite recently switched from an electronic periodic call auction market to an electronic continuous auction market. No doubt it is a rare opportunity for us to deeply look into market quality under different trading mechanisms. Using time-stamped transaction data of trades and quotes covering the period from January 2001 to September 2002, overall behavior for all TX Futures contracts are explored first—including intraday and daily patterns in the bid-ask spreads, volume, trade size, volatility, liquidity ratio and other characteristics. Next, in order to observe whether long-term contracts and short-term contracts have different patterns, the sample is divided into two groups—quarterly expiration contract months (March, June, September, and December) and non-quarterly expiration contract months, and the intraday/daily patterns are displayed. Moreover, since TAIFEX transferred trading mechanism on July 29th 2002 from an electronic periodic call auction market to an electronic continuous auction market, intraday/daily patterns are separately illustrated and compared before and after July 29th 2002, and ANOVA F-Statistic and Kruskal-Wallis tests are also taken to provide more insights into time-varying behavior under two different kinds of market trading mechanisms. The empirical results indicate that the most active periods correspond to the TAIFEX’Ss opening five-minute interval (8:45-8:50), TSEC’Ss opening five-minute interval (9:00-9:05), and TAIFEX’S closing five-minute interval (13:40-13:45) with wide spreads and large trade sizes. In 54 five-minute intervals for the regular trading session of both TAIFEX and TSEC from 9:00 a.m. to 1:30 p.m., the behavior of spreads, volume and trade sizes mainly reveal U-shaped patterns. The average trading volume within each time interval plunges, except within the final 5 minutes closing procedure interval, after TAIFEX transferred trading mechanism from an electronic periodic call auction market to an electronic continuous auction market with wider spreads and narrower volatility, in general. Moreover, intraday patterns of the average volume under new microstructure exhibit a right angular U-shape while intraday patterns of volume under old microstructure reveal a smooth U-shape. The evidence suggests a conjecture that the transfer of market trading mechanism might result in informed traders’ altering their intraday behavior and might lessening their trading desire. Further evidence in confirmation of this statement is left to future work.

台灣期貨市場收盤價的操縱現象並不明顯，推論可能的理由為收盤時採用不連續的搓合機制，使有意操縱收盤價的交易者的操縱意願降低。報酬與操縱呈現負相關，可能的原因為操縱者在不同市場間進行操縱，即使在某一市場虧損，仍能在另一市場獲得更多的報酬。激發操縱交易行為的因子最顯著的解釋變數為交易者的平均累積部位成本，而非收盤價的變動。 / This paper uncovers the trade-based manipulative trading in TAIFEX with the detailed trade-level data. The manipulation of closing prices is rather unremarkable. The call auction may be the reason. Revenue and manipulation have a negative remarkable relation. One of the reasons could be that a manipulator trades between different markets at the same time. He can ear more return in one market to cover the loss in another market. The other possible reason is that the informed trader wants to mislead other traders. This paper finds out that the most explainable motivation of manipulation is the cost of the held position. The outcome supports the negative relation between revenue and manipulation.